Conference Agenda

Abstract submission pending by 31/05/2021

Complex agricultural systems comprise a large array of biophysical and physiological processes so that novel approaches and algorithms are needed to optimize the contribution of Earth Observation (EO) data to the monitoring capability with the aim of developing more sustainable management practices. This paper presents the results of research activities carried out within the ESA-MOST Dragon4 framework focused on the topic of agricultural resources. SINO and EU research groups exploited ESA and Third Party Mission data together with the Chinese EO resources in the agricultural domain both considered at regional or local scale. On different test cases, SINO-EU satellite data have been processed to retrieve biophysical variables related to crop vegetation status and to the derivation of proxy variables linked to the water and nitrogen cycling in the cereals agro-ecosystems. The paper encompasses two research avenues: i) retrievals of biophysical variables of crop and yield prediction; ii) food security. Firstly, optimal procedures (spectral indices, non-kernel-based and kernel-based Machine Learning Regression Algorithms) for retrieving biophysical crop variables (LAI and pigment) by exploiting the spectral information of current multispectral optical satellite were analyzed. Secondly assimilation of multivariate and multi-scale remotely sensed variables into crop models have been performed to retrieve yield and quality estimation under abiotic stress factors both at farm and regional scale. Lastly, field studies allowed to derives spectral index sensitive to the biotic stress (pathogen) of crops. Pathogen models combined to EO data products aim to estimate the rate of pests and diseases recognition and spreading. The combined results of these activities showed that most of the research objectives in the agricultural domain are interconnected and dependent on knowledge flow between them and that the EO exploitation in Chinese and European agriculture, can promote best practices for environmentally and profitable sustainable production through improving resource use efficiency.

This Dragon 4 project 32194 was to investigate the methodology of combined use of European and Chinese high and Medium Satellite data to assess crop and produce crop maps. This project was composed of two subprojects. One was focusing on the high-resolution satellite data and another was focusing on the medium resolution satellite data. The first subproject was entitled crop mapping with time series of high resolution European and Chinese satellite data. The Sentinel-2 and GF-1(GaoFen or high resolution in English) onboard European and Chinese satellites, respectively, were supposed to use as both have quite similar spectral bands. This subproject aimed to take both advantages of Sentinel-2 and GF-1 data to produce a better and earlier crop mapping. The team was supposed to apply and adapt the crop mapping approach of ESA (European Space Agency) Sent2Agri project to a Chinese site with time series of European and Chinese satellite images. The second subproject was entitled assessing crops with PROBA-V (PRoject for On-Board Autonomy–Vegetation) and FY-3 MERSI (Fengyun, Wind and Cloud in English, Medium Resolution Spectral Imager) data. The PROBA-V and FY3-MERSI both have quite similar channels and their own advantages. The new development of this kind of medium resolution satellite data in Europe and China was providing us an opportunity to investigate the possibility and the potential of using both PROBA-V and FY-3 MERSI Data for the crop assessment for large area. This subproject was going to focus on the crop mapping with both satellite data. The team developed the methods to handle both data and then get the information retrieved. With the implementation of this dragon project, the crop type maps were produced in the irrigated area in Northwest China with the sentinel 2A/B,GF-1,PROBA-V and FY3B-MERSI. The overall accuracies of resulting crop type maps from S-2 and GF-1 reached 94-97% while the overall accuracies from PROBA-V and FY3B-MERSI reached 88%. The methodology for the crop type classification was improved and benefited from the Sent2Agri system.

This project made profound studies on the in-orbit calibration and the product quality traceability of space-borne optical sensors, in-orbit calibration and product generation of microwave sensors, MAXDOAS benchmark observation in eastern China, and ground-based FTIR spectrometer benchmark measurement, the related achievements mainly include the following aspects:

(1) In this project, we have developed an SI-traceable automated radiometric calibration system, and established a whole uncertainty transfer link for field calibration. These achievements have been integrated in the "National high resolution remote sensing comprehensive calibration site", namely Baotou site, one of demonstrated sites of the International Committee on earth observation (CEOS)/Global Radiometric Calibration Network (RadCalNet). Under the framework of RadCalNet project, Baotou site could regularly provide the radiometric calibration standard product together with Europe Space Agency (ESA), and supporting the in-orbit performance evaluation of high-resolution sensors onboard series satellites in China, such as GF, ZY, TH, GJ.

(2) According to the higher requirements on radiometric accuracy of microwave RS sensors in the global meteorological and climate research, this project has developed a radiation benchmark transfer model from ground to satellite, and a traceability model from satellite to ground. These models improve the calibration accuracy of single satellite by 17% and improves the consistency of the sensors between generations by 11%.

(3) In-orbit radiometric calibration algorithm of hyperspectral payload also has been developed, and these channels shorter than 312 nm of TROPOMI have been re-calibrated, so that the fitting residual of SO2 spectrum was reduced from 0.40%～0.92% to 0.07%～0.14%. On basis of this, this project has proposed an SO₂ retrieval method from the UV-VIS hyperspectral sensor onboard GF-5 satellite, and the retrieved result has been validated with 121 sets of weekly average in situ data in 12 months. It is demonstrated that bias for this method is reduced by 41%～123% comparing to the official TROPOMI products published by NASA/ESA, this confirms that the official products of ESA seriously overestimate the concentration of SO₂ in China.

In this project, there are 10 participants with senior titles, 15 intermediate titles and 18 graduate students. Among them, Cheng Liu from the University of the Science and Technology of China won the 16th China Youth Science and Technology Award in 2020; Lingling Ma from the Aerospace Information Research Institute, Chinese Academy of Sciences (CAS), and Marc Bouvet from ESA, won the International Cooperation Partner Award for young scientists of CAS in 2020.